Bye Bye Biometrics: AI Demands Stronger Security Standards | Billy Luedtke, Francis B, Evin M

[Music] ladies and gentlemen thank you for coming to today's first performance here at the whale poool stage today we are starting off with bye-bye Biometrics AI demands stronger security standards by eth Sam casset okay great my name is Mo uh I'm a managing partner with Gom cryptos Capital so I'm one of those uh evil VCS that you hear about uh so I'm super excited because these are some of the mega OG Heavy Hitters and intellectual super Giants of this segment so I'm really excited to have this uh panel uh let's actually go through self- inroduction so we'll just do the round robin two-minute self- inroduction uh starting with Evan GM everyone thank you so much for taking time out of your feels like fairly early mornings here at eth Denver my name is Evan McMullen co-founder of pado ID and this morning or actually yesterday morning we just started to roll out the billions network billions. network is the first human and AI Network built with mobile first verification so you can prove who you are from the safety of your phone we leverage zero knowledge attestations to make it easy for you to prove that you are a person a unique person uh that you live in a certain place you are a certain age have certain traits and preferences um so that you can take your identity with you and use it to leverage benefits and personalized rewards both on and offchain that's awesome okay go ahead hello everyone my name is Billy I've been around for a long time I am the founder and CEO of intuition and our goal is to give all things better intuition in all of their interactions by giving them the information they need when they need it from the sources they trust so we're kind of this semantic data discoverability protocol and a oneliner we've been testing out is Google for AI so you guys can let me know how how how well that resonates after this okay let's keep going hey everyone my name is Francis bwaf from ZK pass we are a privacy Centric protocol built on zero knowledge proofs so just imagine that you're able to share information about yourself such as your age or your credit score without disclosing all of the details so that's essentially what we focus on wonderful okay so what I'd like to do is start with kind of the main topic today which is bye-bye Biometrics right so biometric ID fingerprint face who the heck are you right so to me you can't talk about this topic H without invoking everybody's favorite friend which which is cibil right so I guess the thing that I'm curious about is to get reactions from the gang because everyone knows that one way to get rid of cibil is to go to worldcoin and stare into a giant eyeball uh you know created by uh the dark lord Sauron but like you know why why don't you guys take a crack at at you know what do we do about what what do we do about cibil so Evan why don't you start so much to do about civil Mo um I definitely much a do I I definitely think that there are more scalable approaches than creating databases full of body parts from the you know Lord Saron but perhaps if we if we take a step back and talk about just sort of basic skillability and where we find these challenges obviously many of you have probably been on chain before and may have encountered an instance where you are interacting with a non-human counterparty certainly if you've been on crypto Twitter recently the reply guys are out there in force now about 50% of internet traffic today comes from unaccountable Bots that means you know agents as you might uh as you might refer to them depending on the instance but basically um non-human actors that do not have an accountability mechanism no way to ask who's your daddy take me to your leader um and so with that challenge comes um the uh the adverse instance of having you know multiple parties um colluding uh we have the you know the premise of farming but I don't necessarily think that's always a bad thing I think it is a naturally occurring phenoma um but with that in mind we have to think about how we scale um the ability to identify unique parties and as I mentioned earlier I think proprietary Hardware is too big of a bottleneck to be scalable globally and so on our team we focus on mobile devices as a point of entry and a point of identification interesting uh Billy you sound like you look like you have a take yeah I've got I've got a take I I think that we're entering this kind of like interesting philosophical realm where it's going to be what is a Cil attack like at what threshold of intelligence is a bot or an agent treated as kind of like a unique thing that is actually like maybe not bad like sure proof of human like we might want to know if the thing is human or artificial intelligence but then you enter the philosophical realm of are the AI just us and so like are those actually also human and just kind of like the aggregate human consciousness so I don't think that proof of humanity or biometric scans is going to be enough because like you're going to have hardware intelligences and so what I think you need to kind of combat you know civil attacks is just web of trust things saying things about things you listening to the voices you trust and kind of like figuring things out that way that is a helpful mindset so intuition creates a web of trust so that is a mechanism through which you can kind of start to understand who's up to what uh uh Francis uh ZK past take we look at it from uh the perspective of deep fake as well as um basically being able AI generated sort of U counterfeit of uh a human being so anywhere from your voice to your actual features and recogniz recognition can be uh compromised and multiple accounts can be created in that format so we look at it from a cryptographic perspective and stating that if you replace Cibo with uh you know math essentially then you are able to uh avoid that so uh just to kind of give a quick overview when you really start to scrutinize uh last I check this whole ZK pass construct is Loosely coupled to the ZK TLS construct right where you can actually have things like k YC based identity you know through these uh secure channels right so ZK TLS https SSL you know traditional web 2 authentication and then identity based on that right that some of that stuff is pretty hard right so to me I think the next topic is really why why not Biometrics right because I believe that the thing that uh we're saying about the ZK TLS is that uh cibil generally doesn't have a fingerprint generally right or maybe yet right and you know civil generally won't have a kind of a face ID maybe yet right but the the thing that's interesting about Z TLS is that Cil may not have a government ID maybe yet right so that so these are the kind of things we're reduced to right where we're basically like trying to catch cibil and then I think Billy's point with intuition is that cibil maybe doesn't have a lot of trusted friends yet May you know all of these are yet right so you know cibil will be super interesting when cibil has fingers and thumb prints and face ID and a government ID so that'll be an interesting version of cibil uh aan so I think we're actually getting there a little sooner than you might think um to address your question M why not leverage unique Biometrics as the basis or the anchor of your you know your public identifier in short you can't rotate your thumbs you can't rotate your eyeballs tying all of your data uh where your your private key is immutable and immutably tied to your human person makes for a pretty difficult key rotation strategy now I'm sure there will be surgical approaches in theut can rotate her fingerprints but actually what uh what our team just um released this week is what we call the Deep trust framework or the ability to assign a unique identifier to an instance of an agent and to relate that to a unique human being or a set of unique human beings so this process we call it Kya uh know your agent um but this framework allows you to have deep trust in a specific agent instance so it can represent itself as acting on behalf of you uh and can enter spaces where you have privileges to do so um and so we can add a little bit of you know interesting Nuance to the types of coordination games we can play with our agents when we can confer privileges upon them based on our own identity this gets pretty bad though right I mean the scariness that we're talking about here is you know when cibil gets a hold of your fingerprint right that's a super complex situation because what I'm saying here is this right which is that one of the problems that we have inside of the blockchain space is the problem of physical embodiment right and obviously the wrench attack is probably the most famous in incarnation of a security problem that arises out of the fact that we reside inside of bodies right but the thing that's really interesting is like you know the traditional way of reasoning about security is that uh authentication can be based on things that you know like a password right or things that you know you you have like you know or things like like for example Hardware keys right or things that you are or like a you know any kind of uh private key right if you have if you have the private key then you know but the problem with the private key as we all know is that that cibil can get your private key uh you know so I guess to me if you red if you get rid of the Biometrics then like which I understand I understand why you're getting rid of Biometrics but then I'm just like where where do we go Billy I want to give you a shot you I mean I think I think every piece of data is useful so I think biometric data is useful I don't I don't think you should rely on any s kind of single point of failure where it's like oh I just need the world coin eyeball scan and then we're good we solve we solve everything or I don't just need the fingerprint and then we solve it it's like no no no like the way that things are going to emerge is kind of you know you need the aggregate set of data that like the thing has created and the aggregate set of data created about the thing and then you can reason about the reputation of the thing and I think that that is kind of like the true answer to the Civ problem especially in this world where it's you don't just need proof of unique human you need proof of like valuable contributor to thing like you might have a you might have a swarm of bots that are unhuman but they're performing a bunch of really productive action so you like you need to know the reputation of the thing and the reputation is the aggregate of the data about it and that's why something like intuition is so important because it's like allows you to Traverse all of the like verifiable data about the thing in aggregate I think uh let me throw it to you but uh actually you have a quick response to okay so my retort here is it's not a retort actually it's support which is that part of what's happened in the entire crypto space is actually that there's three games that are being played right the three three games are basically the Zero Sum game right which is mostly Traders right like I win you lose right then the positive some game which is I've built something uh and I'm open sourcing it or I'm giving it to you right that's that's the positive some game right and then there's the negative sum game right which is the negative SU game is basically like ransomware like recently there were these huge chat logs that were leaked from a ransomware club and it turned out that one of the arguments inside of the club that they were kind of having was a group of them had started getting the bribe the ransomware money and then not providing the decryption key like basically and and there was kind of this weird ethical argument inside of their chat room where they were like is that cool like can I just rug these guys you know keep their data and get the money right so so that's negative sum right so I'm just describing these three Realms and I just want to say with respect to what you know Billy's talking about he's really talking about how does the crypto community identify who the positive some actors are so they can choose to collaborate with people who are non non malevolent whether they're civil or not right so it's sort of like if you have a benevolent civil then Team Up by all means uh Francis well I was going to add actually uh to to your point and to Billy is even even what we're talking about is not perfect right as as he stated you don't want to eliminate every single uh method that we have instead you want to continue to augment and improve it snarks snarks are not perfect because you need a trusted setup and we all know that's also a point of vulnerability to a degree right if someone has your private they can also manipulate that I mean that's we're not it's not perfect so we we continue even from our side we just continue to search and we used to for just example not getting into any technical terminology here but we used to use Voli based Z you know ZK right and then we combined that with ZK snarks but we saw that that just it wasn't perfect there is also a case where we're not going to be perfect so we just eliminated this ARS automatically and went into this Vol volley in the head which basically does not require trusted setups and even everything VI uh that's verified is publicly out there so it does doesn't have any like private verifiers and so for I mean in this kind of methodology my point is that we just continue to improve and that's all that we can we could do whether you combine biometric with Zer knowledge proofs in order to hide the keys but ensure that uh you know access is still granted to the individual or you eliminate one or the other you have to just make sure that the the tech can speak and that's why we I mentioned earlier that mathematical process is probably better than physical human uh vulnerability and with AI advancing these days I like this I think the the missing element with respect to key rotation of fingerprints is the thing that's interesting is is that you're all you have actually if you count your thumbs as fingerprints you actually have 10 of them them right but the problem is actually key deletion right cuz there isn't provable deletion right so that's a problem right because if you register your index finger right even if you tell your local device that you delete the index finger and you go to the next finger right you're still kind of hosed if like if it's already been authorized in a cloud somewhere right so that that's that that is that's bad but I I still I still feel like I'm still very attached to my fingerprints for some reason you know and and I feel like you're going to pry you're going to have to pry Biometrics out of my cold dead fingers you know maybe it's kind of gross but one I think also maybe this is a little side tangent but one one really important thing to think about is you know like a a fingerprint is okay for maybe like authenticating into your phone and you know maybe they can see your text messages or whatever but if that controls your entire identity and all your money like the the Collision rate isn't zero so that's really not that good and we looked into this uport at uport like I don't know eight years ago and there was like all these companies trying to figure out like things with like Z almost almost zero Collision rate there are companies doing like heartbeat stuff and and all this crazy thing all these crazy things and no one like solved that so if it like you can't I think and like an aggregate set of Biometrics might get you might get you there or an aggregate set of multiple data points but I don't think one biometric thing can really secure your whole life I like that and and also M all I need is your coffee cup in order to get more 10 versions of your fingerprint I just need your coffee cup right now you know that's true I don't you don't have to be dead for me to get your fingerprint yeah but you know I I actually really maintain very tight custody opsc around my coffee cup so like you know you gotta watch out uh so to me the thing that is really I guess the question becomes like where does this go in the future right because obviously we're really talking about kind of like the present right which is you know L we're we're really getting rid of Biometrics right but in the present but like where does this go in the future uh so so I have kind of like a a three-phase view of how the evolution of of data sharing and observability is going down for a long time kind of phase one which I think we've moved out of the government had a monopoly on surveillance the sort of government panopticon the ability to discover and um and you know compute upon or or pursue information about you know about various citizens but that right was not also similarly made available to private Enterprise now with technology we are living in the public private panopticon let's say with a combination of surveillance from our devices cameras on Teslas public CCTV uh you know private security various different mesh networks that are are kind of these overlapping sets of coverage um of data uh that can be compelled by some parties sometimes by individual users but not not always by those individuals and then I think we are moving toward this future state that you might call the personal panopticon where I think that there will be um a broad a much broader variety of uh of methods of data collection and data aggregation about who you are what you do how you've impacted the space around you as Billy was discussing sort of filling out the data required to build a web of trust um and I think that that will uh create an abundant set of data such that we will be able to deduce a shared truth so uh for those of you who are uh looking for some entertainment go ahead and read the Wikipedia page for panopticon it's a very interesting form of prison that actually uh has sort it's frequently cited in uh postmodern or post-st structuralist academic liter so pan panopticon there's your uh New York Times word of the day uh future oh man like I'm I'm I'm just a really big believer in the web of trust cuz that's how like everything works in the world like everything is just social consensus all the way down you have you have like intra subjective consensus where you go around the world and you collect data points and you come to consensus about things being true it's like oh if I jump and I come back down it happens every time so there's like a thing like gravity and then we all observe it and then if if if You observe something that everyone else isn't observing then you're you're crazy so like everything is social consensus so I think like web of trust is kind of the way that the fabric of society and like the I don't know it's kind of the way the world works right now so in order to get there in this kind of like self- Sovereign identity self- Sovereign data or like agent Centric data world where the where the where the entity is the ful Point around which you aggregate the data you need the ability to kind of Traverse all of that data which gets hard if if it's self- Sovereign and it's private so it's like how do you Traverse that whole data set um so you need like discoverability protocols to be able to do that and to request the data you need to authenticate a thing like I I think it's it instead of it being like Oh you know the the counterparty requesting a specific credential like I need the US passport from the US government or I need that plus an eyeball scan from worldcoin it's like what data do you have like what can you present to me and that and and the selective disclosure model of that is is hard when you don't have like a discoverability protocol to understand like what data points you can request so I think like that's the way that things are going to emerge and I think it's just going to be very fluid and it's just going to be things requesting different data points from different things to get them to authenticate or do things or to interact generally I definitely appreciate that my quick RoR is more just as I'm detecting a fundamentally post structuralist overtone right which is in the absence of a dominant Mena narrative we essentially are reduced to sort of like this idea of like collecting a pasti right we're basically collecting whatever we can collect and we're basically building these pastiches so I think that makes sense uh Francis so everything that we're talking about revolving around data and privacy is the only reason this is all important is because it has to do with our own private data we all know that there is more important private data than there is public data out there so if we look into where the world is heading with AI of course uh being at the Forefront then we have to assume that it's not just about my personal information how do I protect it what is the best it's also going to be how I interact externally with ai's involvement so what we looked at and actually we've been researching this for some time is let's look at it from a perspective of these agents and what they're going to help us do in our lives well traditional llms right now they're just out there getting answers and information that's publicly public knowledge out there right but then you have hallucination you have knowledge gaps so instead no one really is looking at private data agents because in this case they cannot get uh information that is let's say specialized medical data data or your own personal bank information in order to provide accurate uh answers nor are they able to uh in this case um basically have uh the most up-to-date information they have to be up to the level where the information was provided to it so we're looking at in inance instead of having coordinator the agents where you as an individual have all your private accounts but you can have your private agent be able to actually accurately get information that you need in order to give you the accurate answers while protecting all of that so you can even share information but without disclosing so much let's say if Evan OB Billy wants something and I have it but it is gated then they can their agent can get that information securely because it is either my research or something that I provided so this is where I see our privacy going Beyond just us but also communicating coordinating with the outside world and individuals and I think the key words uh that Francis uh said really is what I have provided right which is and I think the other word that makes sense in this context is Agent right because agent it has too many meanings and in I think web three agent doesn't mean anything at all because it's abused but I would say that like agent in the sense of representing me right like this speaks a little bit to what Billy was saying earlier about this idea of what what is a person right and if a person expresses their intent to an agent the agent's representing the person then it's kind of that agent kind of of is the person for that for that use case right like that the agent is booking the travel for the person you know it's like it you it may may as well be the person right so you shouldn't think of all sibles as as evil or malevolent right so totally agree with this kind of take I guess another word that popped out in your narrative though was privacy right which I think is it's a word that also has gotten so utterly abused by web 3 that it doesn't mean anything either and it generally means something bad right it generally to me like the word privacy in web 3 kind of means some sort of unsustainable like hyper libertarian construct you know almost like the phrase Free Speech right which is which to me is kind of it doesn't work so like I guess to me the thing that I think is the highest order bit and one of the most genius things that blockchain did is it created the seed of secrecy right which is that we now have a nucleus which is the private key right which is essentially secret data and protected data and we've and we've created a game in which everyone has to have this nucleus right so I guess to me the highest order bid isn't privacy it's actually secrecy and secrecy is interesting privacy is not interesting so you know hot takes so I mean I think I agree with you that privacy gets totally overused I think what's interesting about this concept of uh secret information information known only to you is that we now have uh one Bound for the disclosure of data and then the other is putting put it on chain and clear text the most expressive and public form of you know data distribution that Humanity has ever known I think of privacy as just the absence of disclosure and so um you know it is the end state that arrives after a user gives consent or consent is VI but but I want to retort right because to me the Paradigm of secrecy occurs to groups and organizations so when you think about that then it becomes more like the access control Paradigm rather than this kind of weird political paradigm of of you know absolute privacy whatever that means so uh Billy couple thoughts so you you mentioned one thing like the the the private key kind of being the nucleus around which everything happens es especially kind of the aggregation of data and I think in like what especially in identity world I think people have been thinking about decentralized identifiers for for people most of the time but I think we got to give everything in the world a decentralize identifier like even like every single concept like words and every single product like every single thing in the world should have a decentralized identifier and even like higher order complexity structures of compositions of these different concepts should also have a decentralized identifier and then I want to challenge Billy now to a duel which is the problem with this construct of taking entities and assigning decentralizes identifiers is basically the problem of sign signifier slippage so I mentioned that post structural is and postmodern thought right so the concept of having a symbolic uh unique identifier it only can acre to a symbol because the thing itself is actually like ineffable or intang how's it going to sign yeah so sign signifier slippage is a fundamental uh problem with the unique identification of everything well well like it I think it's very useful to have an identifier it doesn't need to sign it doesn't need to be like a concept you you could think of some crazy stuff where it's like okay maybe like the shareholders of the word trustworthy can all come together to sign on behalf of the I don't know that's a crazy world but there's a lot of things that you can just kind of like you know you could like even counter in counter in counterfactually instantiate like using Create too a smart contract wallet for a thing that never gets claimed by anyone it's just owned by no one and so then the thing has a did that's an eth address but no one nothing can sign with it it just exists so then you can use it as the fulcrum Point around which you aggregate the dat but yes yeah I think that really like that gets into this Beyond of uh the acrel of identity Hood right and it gets into the realm of things like paganism so I think that's a separate panel Francis I don't know where to come what to do but after that kind of interaction I think I think it's better to just say you know secrecy as far as your information should just stay with you and you should choose what to disclose uh you know we used to have the secret see as far as organization like peer reviewers and things of that nature and now we can kind of disseminate it but still sort of have an access point where information could be shared uh but also protected so that's that's my only two cents on that part all right well we're down to the short straws panel wise we have very little time I think we have time for maybe one last kind of hot take and it really kind of for me it has to do with sort of authentication because that's the base right which is if you don't have Authentication can't assign an identifier and you can't know what to do everything's built on top of that so I guess I guess any last hot takes on authentication should not be predicated on a single unchangeable trait um definitely plus one to Billy's note on the web of trust superseding lists and multiactor immovable factors and and multiactor authentication yeah any any other hot takes hot take is I think kind of if you look at the scheme of like a public private key pair like having something that's public and something that's private is really useful so I think in this world of like private self- Sovereign data you kind of want to reveal like a little bit of data so that people can find it and request it cuz if you if everything's secret you don't even know what to request so you got to like reveal a little bit of stuff so people can know what to request from you or what how where to find things final seconds Francis any last comment care about privacy I like it I think it's good that's a the Privacy word pops up all right well well I'm just going to close with this idea that multiactor authentication is here you don't really have to get rid oh they cut me off good job thank you very much to bye biometric thank you Mo Billy Francis and